# scGestaltPrepFunc
# -------------------------------------------------------------------------------------------
# The following functions will handle pre-processing of inDrops scGestalt data 
# prior to running the scGestalt analysis pipeline (if necessary)
# -------------------------------------------------------------------------------------------

# Starting with a single inDrops sequence fastq file containing reads for all inDrops cell index identifiers, 
# split the fastq into individual fastq files for each cell.
# Cell index identifier will be in the header for each read


SplitReadsByBarcode = function(read.dat, filepath, sample.id) {
  #   read.dat = vector with fasta sequence
  #   filepath = full path name for output directory 
  #   sample.id = sample prefix for file name
  #   Outputs directory with as many files as cell index identifiers
  #   Structure of filename: SampleId_Barcode.fq
  
  i=1
  while(i<length(read.dat)) {
    minifile = read.dat[i:(i+3)]
    header.split = strsplit(minifile[1], ":")
    barcode = substr(header.split[[1]][1], 2, nchar(header.split[[1]][1]))
    file.name = paste(filepath, "/", sample.id, "_", barcode, ".fq", sep="")
    
    if(!dir.exists(filepath)) {
      dir.create(filepath)
    }
    if(!file.exists(file.name)) {
      writeLines(minifile, file.name)
    } else {
      cat(minifile, file = file.name, append = TRUE, sep="\n")
    }
    i=i+4
  }
}




# Input directory has a subdirectory with list of files that have been split according 
# to the cell index identifier, such that each file has all the reads for a given cell.
#
# Input directory also has a copy of the fastq file (umi.dat) with random 10base "UMI sequences"
# to be prepended to each read of a cell (this is such that the scGESTALT pipeline can be run with the "UMI" parameter).
# umi.dat is provided as a file under the TestData folder in Supplementary Software 1 scGESTALTAnalysisPipe.
# Different cells will have distinct "UMIs" prepended to their reads.
# Alternatively, these steps can be SKIPPED when the UMI parameter on the scGESTALT analysis pipeline is turned off.


UMISplit = function(umi.dat) {
  #   Split an initial umi.dat file into a list of length = number of cells,
  #   where each entry of the list is a vector of size 4 containing UMI fastq values.
  #   Each cell will be assigned a distinct "UMI barcode" that will represent its cell ID.
  #
  #   umi.dat = fastq file of length = 4 x number of cells, contains distinct UMIs 
  
  i = 1
  j = 1
  numiteration = as.list(1:(length(umi.dat)/4)) # define the length of the list = (length of fastq)/4
  while (j <= length(umi.dat)) {
    numiteration[[i]] = umi.dat[j:(j+3)]
    i = i+1
    j = j+4
  }
  return(numiteration)
}



PrependFastq = function(cell.dat, umi.dat, cell.id) {
  #   Prepend a given "UMI" to each read of a given cell
  # 
  #   cell.dat = list of cells where each list subset contains all fastq reads for that cell
  #   umi.dat = list of umis where each list subset contains the umi fastq reads which will be
  #   assigned to each read of a cell
  #   cell.id = prefix for file name
  
  a = 1 # counter for cell.dat and umi.dat list (note lengths of both are same by definition)
  list.names = vector()
  while (a <= length(cell.dat)) { # number of iterations = size of cell.dat & umi.dat lists
    temp = length(cell.dat[[a]]) 
    i = 2
    j = 4
    while (i <= temp) { # num of iteration = num of reads per cell for each cell subset of the cell.dat list
      cell.dat[[a]][i] = paste(umi.dat[[a]][2],cell.dat[[a]][i], sep="") # prepend UMI sequence
      cell.dat[[a]][j] = paste(umi.dat[[a]][4],cell.dat[[a]][j], sep="") # prepend FASTQ score
      i = i + 4
      j = j + 4
    }
    list.names[a]= paste(cell.id[a], "_cell_", a, "_", umi.dat[[a]][2], sep="")
    a = a + 1
  }
  names(cell.dat) = list.names
  return(cell.dat)
}